Photoswitchable dynamic combinatorial libraries: Coupling azobenzene photoisomerization with hydrazone exchange

Article abstract of DOI:10.1021/jo801783w

(Chemical Equation Presented) The novel azobenzene-based monomer 1 was prepared, equipped with the necessary functionality to undergo simultaneous dynamic exchange processes: hydrazone exchange and photoisomerization. Acid-promoted hydrolysis of the azobe

Full text of DOI:10.1021/jo801783w

Photoswitchable Dynamic Combinatorial Libraries: Coupling
Azobenzene Photoisomerization with Hydrazone Exchange
Lindsey A. Ingerman and Marcey L. Waters*
Department of Chemistry, CB 3290, UniVersity of North Carolina, Chapel Hill, North Carolina 27599
mlwaters@email.unc.edu
ReceiVed August 8, 2008
The novel azobenzene-based monomer 1 was prepared, equipped with the necessary functionality to
undergo simultaneous dynamic exchange processes: hydrazone exchange and photoisomerization. Acid-
promoted hydrolysis of the azobenzene building block produced a dynamic combinatorial library of cyclic
oligomers, while multibuilding block libraries were also generated upon addition of proline-based
monomers. Libraries equilibrated under thermal conditions were dominated by trans isomers of the
azobenzene macrocycles, whereas light-induced isomerization resulted in a conformational change of
the library members to their corresponding cis-azo form. In the presence of a pentaproline template, a
slower rate of thermal relaxation of the cis-azobenzene species 1c was observed, resulting in stabilization
and amplification of this receptor due to favorable binding interactions. The facile identification and
application of such photoswitchable receptors have the potential to allow for greater control over molecular
recognition events.
Introduction
Dynamic combinatorial chemistry (DCC) is an attractive
alternative to rational design in the development of novel
receptors in which molecular recognition guides the synthesis
of complex host systems from simple building blocks using
reversible linkages under thermodynamic control (Figure 1).1
It takes advantage of the ability of a molecular target to template
the preferential bond formation of the strongest target binders,
upon which desired receptors are amplified at the expense of
other oligomers.2 DCC not only allows for the discovery of
nonintuitive receptors, but it also provides access to structures
that may otherwise be unobtainable by traditional synthesis.3
FIGURE 1. Generation and templation of a dynamic combinatorial
library (DCL).
In recent years, DCC has found application among a variety of
disciplines, ranging from drug discovery to material science.4
Incorporating building blocks with new features and poten-
tially interesting recognition elements is desirable in order to
(1) For reviews, see: (a) Corbett, P. T.; Leclaire, J.; Vial, L.; West, K. R.;
Wietor, J.-L.; Sanders, J. K. M.; Otto, S. Chem. ReV. 2006, 106, 3652–3711. (b)
Lehn, J.-M. Chem.sEur. J. 1999, 5, 2455–2463. (c) Ladame, S. Org. Biomol.
Chem. 2008, 6, 219–226. (d) Ludlow, R. F.; Otto, S. Chem. Soc. ReV. 2008, 37,
101–108. (e) Otto, S. Curr. Opin. Drug Disc. DeVel. 2003, 6, 509–520.
(2) For example, see: (a) Otto, S.; Furlan, R. L. E.; Sanders, J. K. M. Science
2002, 297, 590–593. (b) Baolu, S.; Stevenson, R.; Campopiano, D. J.; Greaney,
M. F. J. Am. Chem. Soc. 2006, 128, 8459–8467. (c) Voshell, S. M.; Lee, S. J.;
Gagne, M. R. J. Am. Chem. Soc. 2006, 128, 12422–12423.
(3) (a) Lam, R. T. S.; Belenguer, A.; Roberts, S. L.; Naumann, C.; Jarrosson,
T.; Otto, S.; Sanders, J. K. M. Science 2005, 308, 667–669. (b) Hamilton, D. G.;
Feeder, N.; Teat, S. J.; Sanders, J. K. M. New J. Chem. 1998, 101, 9–1021.
(4) (a) Otto, S.; Furlan, R. L. E.; Sanders, J. K. M. Drug DiscoV. Today
2002, 7, 117–125. (b) Karan, C.; Miller, B. L. Drug DiscoV. Today 2000, 5,
67–75. (c) Lehn, J.-M. Prog. Polym. Sci. 2005, 30, 814–831.
10.1021/jo801783w CCC: $40.75
Published on Web 11/25/2008
2009 American Chemical Society
J. Org. Chem. 2009, 74, 111–117 111

Ingerman and Waters
thermally, although thermal relaxation to the trans state is a
slow process (hour-to-day time scale).10
The conformational change that is induced upon isomerization
of azobenezene derivatives has been successfully exploited to
control the biological properties of various systems, such as
folded peptides11 and helical polymers,12 by either disrupting,
changing, or enhancing secondary structure. Recently, progress
has been made in the development of photoswitches that
covalently modify target proteins and reversibly present and
withdraw a ligand from its binding site as a result of photoi-
somerization of an azobenzene linker, allowing for rapid and
selective manipulation of protein function.13 The photoswitch-
able properties of azobenzenes have also been utilized to
manipulate the properties of host-guest systems involving
crown ethers14 and cyclodextrins,15 while also finding applica-
tions as small molecule inhibitors.16
FIGURE 2. Cis-trans isomerization of azobenzene derivatives.
expand the present applications of DCC. In particular, several
groups have developed “doubly dynamic” DCLs through the
incorporation of two different reversible reactions which can
be triggered independently, demonstrating the benefit of incor-
porating multiple equilibria in a single system.5 For example,
Otto and co-workers have prepared DCLs which feature two
simultaneous covalent exchange reactions, disulfide and thioester
exchange.6 In this system, the two reversible reactions are
addressed sequentially. First, library equilibration occurs based
on thioester exchange only, and then in the presence of
atmospheric oxygen both reactions occur simultaneously. Eliseev
and co-workers have explored doubly dynamic DCLs as well
by combining noncovalent metal coordination with imine
exchange, which can be used as independent equilibrium
processes controlled by different types of external intervention,
oxidation/reduction of the metal template, and change in the
pH and temperature of the medium.7 Furthermore, a system has
been reported in which three dynamic linkages, disulfide, imine,
and coordinative bonds, were shown to be capable of simulta-
neous reversible exchange.8 Although the three types of dynamic
linkages were demonstrated to be mutually compatible, both
transmetalation and covalent imine exchange were used to alter
the equilibrium between disulfides, allowing for greater control
over the degree of self-sorting.
Herein, we report a further expansion in the area of multilevel
dynamic libraries by combining two reversible processes,
hydrazone exchange and photoinduced isomerization. The two
exchange processes involved in our double-level DCLs is
advantageous in that it offers a higher degree of control over
the library composition in the investigation of potential targets.
While hydrazone exchange facilitates the traditional formation
and interconversion of an assembly of macrocycles under acidic
conditions, photoinduced isomerization can be applied for the
development of switchable receptors. We aimed to develop a
DCL from which we could identify switchable receptors with
which one could photomodulate molecular recognition processes
as a direct result of the distinct conformational changes of
azobenzene.
Despite its extensive use in other applications, particularly
in the field of molecular recognition, photochemistry has yet to
be widely employed in the design of dynamic combinatorial
libraries. Eliseev and co-workers reported an early example
which integrated photoisomerization into DCC, making use of
an unsaturated dicarboxylate monomer in the development of
anionic receptors for arginine.17
Hydrazone exchange is well suited for DCC, as much success
has been met with this reversible reaction in the preparation of
DCLs,18 and the required functionality can be incorporated into
an azobenzene derivative in a straightforward manner. The
hydrazone linkage is formed from a hydrazide and an aldehyde
under acidic conditions. While acid catalyzes both the initial
formation and the interconversion of an assembly of macro-
cycles, neutralization yields stable, isolable products.19
These two types of exchange were merged in the design of
a novel azobenzene-containing building block appended with
the appropriate functionality for hydrazone exchange. We have
investigated the generation of various macrocycles incorporating
our designed azobenzene building block 1 via DCC (Figure 3),
examining both single building block libraries and libraries with
more than one building block, adding to the structural diversity
of the library. The additional building blocks include the proline-
based monomers 2 and 3.20 We report the composition of DCLs
derived from the azobenzene building block 1 as a single
(10) Kumar, G. S.; Neckers, D. C. Chem. ReV. 1989, 89, 1915–1925.
(11) For example, see: (a) Renner, C.; Kusebauch, U.; Loweneck, M.;
Milbradt, A. G.; Moroder, L. J. Peptide Res. 2005, 65, 4–14. (b) Pieroni, O.;
Fissi, A.; Angelini, N.; Lenci, F. Acc. Chem. Res. 2001, 34, 9–17. (c) Jurt, S.;
Aemissegger, A.; Guntert, P.; Zerbe, O.; Hilvert, D. Angew. Chem., Int. Ed.
2006, 45, 6297–6300.
(12) For a review, see: (a) Natansohn, A.; Rochon, P. Chem. ReV. 2002,
102, 4139–4175. (b) Iftime, G.; Labarthet, F. L.; Natansohn, A.; Rochon, P.
J. Am. Chem. Soc. 2000, 122, 12646–12650. (c) Tie, C.; Gallucci, J. C.; Parquette,
J. R. J. Am. Chem. Soc. 2006, 128, 1162–1171.
(13) Harvey, J. H.; Trauner, D. ChemBioChem 2008, 9, 191–193.
(14) Shinkai, S.; Manabe, O. Top. Curr. Chem. 1984, 121, 67–104.
(15) Ueno, A.; Tomita, Y.; Osa, T. Tetrahedron Lett. 1983, 24, 5245–5248.
(16) Abell, A. D.; Jones, M. A.; Neffe, A. T.; Aitken, S. G.; Cain, T. P.;
Payne, R. J.; McNabb, S. B.; Coxon, J. M.; Stuart, B. G.; Pearson, D.; Lee,
H. Y.-Y.; Morton, J. D. J. Med. Chem. 2007, 50, 2916–2920.
(17) Eliseev, A. V.; Nelen, M. I. J. Am. Chem. Soc. 1997, 119, 1147–1148.
(18) For example, see: (a) Matsumoto, M.; Nicholas, K. M. J. Org. Chem.
2007, 72, 9308–9313. (b) Liu, J.; West, K. R.; Bondy, C. R.; Sanders, J. K. M.
Org. Biomol. Chem. 2007, 5, 778–786. (c) Simpson, M. G.; Watson, S. P.; Feeder,
N.; Davies, J. E.; Sanders, J. K. M. Org. Lett. 2000, 2, 1435–1438. (d) Roberts,
S. L.; Furlan, R. L. E.; Otto, S.; Sanders, J. K. M. Org. Biomol. Chem. 2003, 1,
1625–1633.
Azobenzene has been widely used as an optical trigger for
various photoresponsive systems due to its pronounced changes
in geometry upon light-induced isomerization. Azobenzene is
an attractive photoswitch due to its high phototstability, facile
isomerization resulting in good quantum yields, and extremely
fast and reversible isomerization processes (picosecond time
scale).9 At thermal equilibrium, the trans isomer is dominant,
but irradiation to the photostationary state converts the trans
isomer to its corresponding cis form (Figure 2). The reverse
process is also feasible photochemically (at 450 nm) or
(5) (a) Miller, B. L.; Klekota, B. Tetrahedron 1999, 55, 11687–11697. (b)
ten Cate, A. T.; Dankers, P. Y. W.; Sijbesma, R. P.; Meijer, E. W. J. Org. Chem.
2005, 70, 5799–5803.
(6) Leclaire, J.; Vial, L.; Otto, S.; Sanders, J. K. M. Chem. Commun. 2005,
1959–1961.
(7) Goral, V.; Nelen, M. I.; Eliseev, A. V.; Lehn, J.-M. Proc. Nat. Acad.
Sci. U.S.A. 2001, 98, 1347–1352.
(8) Sarma, R. J.; Otto, S.; Nitschke, J. R. Chem.sEur. J. 2007, 13, 9542–
9546.
(19) Cousins, G. R. L.; Poulsen, S.-A.; Sanders, J. K. M. Chem. Commun.
1999, 1575–1576.
(20) Furlan, R. L. E.; Ng, Y.-F.; Cousins, G. R. L.; Redman, J. E.; Sanders,
J. K. M. Tetrahedron 2002, 58, 771–778.
(9) Renner, C.; Moroder, L. ChemBioChem 2006, 7, 868–878.
112 J. Org. Chem. Vol. 74, No. 1, 2009

Photoswitchable Dynamic Combinatorial Libraries
SCHEME 1. Synthesis of Building Block 1a
FIGURE 3. Azobenzene building block 1 and proline building blocks
2 and 3 functionalized with a hydrazide (red) and protected aldehyde
(blue) to facilitate hydrazone exchange.
component and in mixtures, along with the effect of photoi-
somerization on the composition of such libraries.
Results and Discussion
Synthesis of Building Blocks and Preparation of DCLs.
The azobenzene building block 1 was synthesized from the
corresponding fully protected azobenzene amino acid 4 (Scheme
1). The preparation of 4 relies on the reaction of a nitrosoben-
zene with an aniline as described by Hilvert and co-workers.21
The nitrosobenzene was prepared in two steps from com-
mercially available m-nitrophenylacetic acid, whereas the aniline
was prepared in two steps from commercially available 3-ni-
trobenzylamine hydrochloride. The protected amino acid 4 was
transesterified to the corresponding methyl ester 5, followed by
Fmoc deprotection under basic conditions, to yield the free
amine 6.22 The coupling of 6 with 3-carboxybenzaldehyde
dimethoxy acetal via standard HOBt/HBTU coupling to yield
7 followed by hydrazinolysis of the methyl ester afforded the
desired building block 1 as a mixture of cis and trans isomers,
containing the necessary functionality to facilitate hydrazone
exchange.
a Conditions: (a) H2SO4, MeOH (94%); (b) TAEA, CH2Cl2 (68%); (c)
3-carboxybenzaldehyde dimethoxy acetal, HOBt, HBTU, TEA, DMF (60%);
(d) NH2NH2 ·H2O, MeOH (65%).
Deprotection and subsequent cyclization of 1 (4 mM) was
accomplished using an excess of TFA (100 mM). Because we
were met with solubility problems upon cyclization in most
solvents, libraries prepared with building block 1 alone were
generated in DMSO. Reactions were monitored daily by
LC-MS for 4-5 days, although thermal equilibrium was
reached after 4 days. Most cyclization occurred within 24 h,
and the initial library distribution did not change drastically after
the first day. Although linear species were observed within hours
of monomer deprotection, the libraries were composed of
entirely cyclic macrocycles upon reaching equilibrium. This
DCL gave relatively simple distributions dominated by mac-
rocyclic monomers and dimers, along with smaller trimer peaks
(Figure 4a).
Each library member was observed in both the cis and trans
conformations, and the isomers were easily identified due to
different retention times and a large difference in absorbance
of trans and cis-azobenzene at 320 nm (see Figure S9.
FIGURE 4. HPLC traces at 280 nm of a DCL of 1 (4 mM) (a) at
thermal equilibrium and (b) after isomerization. The chromatogram
y-axes are on the same scale.
Supporting Information).23 The cyclic hydrazones were found
in a cis to trans ratio of about 20:80 at thermal equilibrium. To
ascertain the reversible nature of the library and confirm that
the reaction mixture had reached equilibrium, the library was
also generated from pure cyclic trans monomer. As expected,
the final product distribution was the same (not shown).
(21) (a) Aemissegger, A.; Krautler, V.; van Gunsteren, W. F.; Hilvert, D.
J. Am. Chem. Soc. 2005, 127, 2929–2936. (b) A similar synthesis of the
corresponding free acid of 4 was reported by Renner and co-workers: Dong,
S.-L.; Loweneck, M.; Schrader, T. E.; Schreier, W. J.; Zinth, W.; Moroder, L.;
Renner, C. Chem.sEur. J. 2006, 12, 1114–1120.
(22) (a) Unable to prepare methyl ester 5 directly due to the instability of
the methyl ester of 4-nitrosobenzoic acid. (b) Ulysse, L.; Chmielewski, J. Bioorg.
Med. Chem. Lett. 1994, 4, 2145–2146.
(23) Griffiths, J. Chem. Soc. ReV. 1972, 1, 481–493.
J. Org. Chem. Vol. 74, No. 1, 2009 113

Ingerman and Waters
FIGURE 5. HPLC traces at 280 nm of a DCL of 1 (4 mM) and 2 (4
mM) (a) at equilibrium and (b) immediately after photoisomerization.
The chromatogram y-axes are on the same scale.
FIGURE 7. Part of the HPLC traces at 280 nm of a DCL of 1 (4 mM)
and 3 (4 mM) (a) immediately after irradiation of an equilibrated library,
(b) 6 days after irradiation where there has been adequate time for
thermal relaxation, and (c) equilibrated in the photostationary state with
repetitive isomerization. The chromatogram y-axes are on the same
scale.
libraries. Upon irradiation of equilibrated DCLs at 365 nm for
5 min, each library distribution is shifted significantly in favor
of the cis isomers. This assisted in confirming our assignments
of the different isomers. Further isomerization does not occur
with longer light exposure. After irradiation of the single
building block library, the azobenzene cyclic monomer species
is present in a cis:trans ratio of 87:13 as observed by LC-MS
(Figure 4b). In this library, a similar large increase of the cis/
cis azobenzene homodimer (1c ·1c) is observed, accompanied
by a significant decrease of the trans/trans azobenzene ho-
modimer (1t ·1t) and a minor decrease in the cis/trans het-
erodimer (1c ·1t). Comparable changes are also observed with
the trimer macrocycles (1·1·1), as well as with oligomers
containing one or more proline monomer units in the multi-
building block libraries (Figures 5b and 6b).
Immediately after photoisomerization, the libraries are stored
in the dark, although with time they return to thermal equilib-
rium. The rate of conversion back to thermal equilibrium seems
to be solvent dependent, taking on the order of a week or less
for the libraries generated in mostly CHCl3 but longer for those
in DMSO only. Hydrazone exchange within the irradiated DCL
(Figure 7a) and relaxation back to thermal equilibrium are
competitive processes, occurring simultaneously, and resulting
in a DCL dominated by trans macrocycles (Figure 7b). In
contrast, libraries which are equilibrated under photochemical
conditions with repetitive irradiation allow for hydrazone
exchange to occur in the absence of thermal relaxation (Figure
7c). By comparing parts a and c of Figure 7, both in the
photostationary state, it is evident that the order of equilibration
and photoisomerization along with the conditions under which
the DCLs are equilibrated considerably influence the distribution
of macrocycles. The integration of these two dynamic reversible
FIGURE 6. HPLC traces at 280 nm of a DCL of 1 (4 mM) and 3 (4
mM) (a) at equilibrium and (b) immediately after photoisomerization.
The chromatogram y-axes are on the same scale.
Library Diversification. To further diversify the library and
expand upon the number of host macrocycles generated, the
flexible azobenzene building block 1 was reacted with the more
rigid proline-based building blocks 2 and 3, differing only in
the position of the dimethoxyacetal substituent (meta and para,
respectively). The necessary functional groups for hydrazone
exchange were similarly installed in these monomers, following
modified literature procedures.19,20
The DCLs were prepared (4 mM in each monomer) in a
CHCl3-DMSO (85:15 v/v) solution, as the proline monomer
has been found to assist with solubility and CHCl3 is a more
amenable solvent for templation studies. After equilibration,
these multibuilding block DCLs generated complex library
distributions with upward of 20 identifiable species and up to
pentameric cyclic oligomers in some cases (Figures 5a and 6a).
The distinct differences between the distribution of species
in the libraries generated with 2 versus 3 portray the sensitivity
of DCLs to slight structural changes in the building blocks.
Proline monomer 2 is found to self-sort to a much larger extent
than 3, giving rise to the proline homodimer (2·2) as the
dominant library member in DCLs generated from 1 and 2,
whereas the heterodimer (1·3) is found to be the major species
in DCLs generated from 1 and 3.
Effect of Photoisomerization on Library Distribution. To
investigate the dual nature of our libraries and their application
in the development of photoswitchable hosts, we examined the
effect of light-induced isomerization on the composition of our
114 J. Org. Chem. Vol. 74, No. 1, 2009

Photoswitchable Dynamic Combinatorial Libraries
FIGURE 8. Part of the HPLC trace at 280 nm of a DCL made from 1 (4 mM) and 2 (4 mM) (a) at thermal equilibrium without an added template
(b) at thermal equilibrium in the presence of a pentaproline peptide (25 mM) (c) immediately after irradiation, untemplated, (d) immediately after
irradiation with a pentaproline peptide, (e) 3 days after irradiation, untemplated, and (f) 3 days after irradiation with a pentaproline peptide. Note
that (e) and (f) are not meant to represent DCLs which have returned to thermal equilibrium. Amplified species are indicated with the dashed lines
between untemplated and templated libraries. The chromatogram y-axes are on the same scale.
processes permits various distributions of species to be generated
under different conditions within a single DCL. Such versatility
within a library may prove to be valuable in expanding the
applications of DCC, as well as in the identification of
photoswitchable receptors in this context.
Azobenzene monomeric building blocks which were subjected
to photoisomerization before cyclization were also investigated.
As expected, upon acid catalysis the irradiated monomers
underwent simultaneous cyclization and thermal relaxation,
which over time resulted in DCLs identical to those generated
with the thermodynamically favored trans monomers.
Templation Studies. Because photoisomerization converts
the library to a photostationary state rather than a thermally
equilibrated state, if a guest binds a specific cis macrocycle it
would be expected to inhibit conversion of that receptor back
to its trans conformation due to favorable binding interactions.
FIGURE 9. Comparison between the extent of amplification of each
library member in the polyproline templated library (made from 1 and
2) at thermal equilibrium (yellow) and 3 days after irradiation (blue).
(Percent amplification ) [(% area of library member in templated DCL
- % area of library member in untemplated DCL)/% area of library
member in untemplated DCL] × 100).
Although this is not amplification in the traditional sense, it
would nonetheless result in an increased amount of the cis
receptor in the templated library versus the corresponding
untemplated DCL. This is indeed what was observed when the
libraries were equilibrated in the presence of a pentaproline
guest.
We chose to investigate an oligoproline peptide guest due to
the role of polyproline helices in many important protein-protein
interactions.24 When a photoswitchable DCL consisting of
monomers 1 and 2 was thermally equilibrated in the presence
of a five-residue oligoproline peptide, the 1c ·1t heterodimer is
amplified by about 48% relative to the untemplated library
(Figure 8a,b and Figure 9). Immediately following photoisomer-
ization, amplification of both 1c ·1t and 1c ·1c is observed relative
to the untemplated library, yet there has been little time for
requilibration of the newly isomerized macrocycles at this point
(Figure 8c,d).
host-guest binding interactions. This rate difference is most
apparent with the cyclic cis monomer 1c. As seen in parts e
and f of Figure 8 along with Figure 9, 3 days after the DCLs
were irradiated at 365 nm, host 1c is significantly amplified in
the polyproline templated DCL. As expected, this amplification
is observed in DCLs containing monomer 1 and either proline
monomer (2 or 3). In contrast, 1t is not amplified in the thermally
equilibrated library, suggesting that the conformation of the
azobenzene controls binding. The fact that 1c is only amplified
after photoisomerization is likely a result of not only the low
concentration of cis-azobenzene macrocycles at thermal equi-
librium but also the competitive equilibria at play. Presumably
at higher concentrations of the cis monomer, the favorable
interactions which result in templation of 1c out-compete other
equilibria. This type of phenomenon is not unprecedented in
dynamic combinatorial libraries25 and reinforces the fact that
In returning to a state of thermal equilibrium, a slower rate
of thermal relaxation of some library members was observed
in comparison to the untemplated library, indicative of favorable
(24) Ball, L. J.; Kuhne, R.; Schneider-Mergener, J.; Oschkinat, H. Angew.
Chem., Int. Ed. 2005, 44, 2852–2869.
(25) (a) Corbett, P. T.; Sanders, J. K. M.; Otto, S. Angew. Chem., Int. Ed.
2007, 46, 8858–8861. (b) Severin, K. Chem.sEur. J. 2004, 10, 2565–2580.
J. Org. Chem. Vol. 74, No. 1, 2009 115

Ingerman and Waters
the thermally equilibrated and photoisomerized libraries repre-
sent different libraries with different behaviors which indeed
should be considered independently.
Experimental Section
Synthesis. (E)-Methyl 2-(3-((3-((((9H-Fluoren-9-yl)methoxy)-
carbonylamino)methyl)phenyl)diazenyl)phenyl)acetate (5). In 27
mL of methanol, 0.1 M in H2SO4, 0.29 g (0.53 mmol) of 4 was
dissolved, and the solution was refluxed for 6 h. After removal of
the solvent, water was added to precipitate the product, and it was
extracted three times with EtOAc. The organic extracts were dried
over MgSO4, and the solvent was evaporated to obtain an orange
Unfortunately, as is often the case with hydrazone libraries,
the binding constants of the isolated host-guest complex could
not be measured because the host is not soluble under non-
equilibrating (i.e., nonacidic) conditions, except in DMSO,
which disrupts binding.2c,26
1
solid (94%). H NMR (CDCl3, 300 MHz): δ 7.83-7.81 (m, 4H),
The observed stabilization of a cis-azobenzene via binding
is not unprecedented. Such stabilization has been previously
reported in the context of binding tethered, photoswitchable
maleimide-azobenzene-glutamate ligands.27 Polyproline pep-
tides and other such guests which form more favorable binding
interactions with one isomer of a specific host are especially
attractive due to the potential to photomodulate the binding to
these receptors, therefore achieving a larger degree of control
over such molecular recognition processes.
Limitations of the Azobenzene Hydrazide DCLs. Despite
the desirable properties of azobenzenes, based on our initial
studies there are also some limitations with azobenzene building
blocks that must be noted. First, in the presence of nucleophiles
such as halide counterions, the cis-azobenzene isomers are
rapidly converted to their trans counterparts under the acidic
conditions necessary for hydrazone exchange. This clearly
perturbs the equilibrium, yet is unrelated to favorable binding
interactions, and limits the range of guests that can be studied.
Second, the azobenzene moiety can undergo surprisingly facile
reduction to hydrazobenzene in the presence of reducing agents
including thiols such that azobenzene isomerization is not
compatible with disulfide or thioester exchange.28
7.75 (d, J ) 7.5 Hz, 2H), 7.59 (d, J ) 7.5 Hz, 2H), 7.50-7.44 (m,
2H), 7.41-7.36 (m, 4H), 7.31-7.26 (m, 2H), 5.26 (t, J ) 6.0 Hz,
1H), 4.47-4.45 (m, 4H), 4.23 (t, J ) 6.8 Hz, 1H), 3.72 (s, 2H),
3.70 (s, 3H). HRMS-ESI(+): m/z [M + H]+ calcd for C31H27N3O4
506.2080, found 506.2062; m/z [M + Na]+ calcd for C31H27N3NaO4
) 528.1899, found ) 528.1907.
(E)-Methyl 2-(3-((3-(Aminomethyl)phenyl)diazenyl)phenyl)acetate
(6). To a solution of 0.26 g (0.52 mmol) of 5 in 5.2 mL of
dichloromethane was added 3.9 mL (26 mol) of tris(2-aminoethy-
l)amine. The mixture was stirred for 1 h and then washed with
saturated NaCl followed by phosphate buffer pH 5.5. The organic
extracts were dried over MgSO4. After the solvent was evaporated,
the resulting oil was purified by flash chromatography (MeOH with
10% NH4OH/CH2Cl2 1:9) to obtain 0.18 g (68%) of a bright orange
oil. The product was isolated as a mixture cis and trans isomers,
with the cis isomer ranging from 10 to 20% as determined by NMR
1
integrations. H NMR (CDCl3, 300 MHz): δ 7.85-7.77 (m, 4H),
7.49-7.37 (m, 4H), 3.96 (s, 2H), 3.72 (s, 2H), 3.70 (s, 3H). HRMS-
ESI(+): m/z [M + H]+ calculated for C16H17N3O2 ) 284.1399,
found ) 284.1402.
(E)-Methyl 2-(3-((3-((3-(Dimethoxymethyl)benzamido)meth-
yl)phenyl)diazenyl)phenyl)acetate (7). A solution of 0.022 g (0.112
mmol) of 3-carboxybenzaldehyde dimethoxy acetal, 0.015 g (0.112
mmol) of HOBt, and 0.042 g (0.112 mmol) of HBTU in 700 µL of
DMF was added to 0.032 g (0.112 mmol) of 6. To the mixture
was added 47 µL (0.336 mmol) of triethylamine, and the solution
was stirred for 4 h. After the addition of brine, the product was
extracted with EtOAc. The organic phase was washed with sodium
bicarbonate and dried over MgSO4, and the solvent was evaporated.
The resulting orange oil was purified by flash chromatography
(MeOH with 10% NH4OH/CH2Cl2 1:19) to obtain 0.031 g (60%)
of a bright orange oil. The product was again isolated as a similar
mixture of isomers, with a small amount of the corresponding
deprotected aldehyde (5-10%) which was carried through without
Conclusion
This work reports the design of an azobenzene-based hydazide
monomer that can undergo thermodynamically controlled oli-
gomerization and cyclization under acidic conditions, either in
single or multibuilding block libraries. Although the thermally
equilibrated DCLs favor the trans isomers, we have shown that
we can photochemically control the library distribution by
irradiating the systems at appropriate wavelengths. The incor-
poration of these two reversible processes in a single library is
advantageous in that it offers a higher degree of control over
the library composition and its organization in the investigation
of potential targets. In exploring the templation of a polyproline
peptide, we have demonstrated that we can stabilize and amplify
an inherently less stable cis-azobenzene macrocycle through
templation with an appropriate guest. The development of hosts
for such proline rich sequences via DCC may help to elucidate
their role as potential protein interaction domains.
1
a problem. H NMR (CDCl3, 300 MHz): δ 7.86-7.80 (m, 6H),
7.57 (d, J ) 7.5 Hz, 1H), 7.48-7.37 (m, 5H), 5.39 (s, 1H), 4.72
(d, J ) 5.7 Hz, 2H), 3.71 (s, 2H), 3.69 (s, 3H), 3.29 (s, 6H). HRMS-
ESI(+): m/z [M + Na]+ calcd for C26H27N3NaO5 ) 484.1849, found
) 484.1861.
(E)-3-(Dimethoxymethyl)-N-(3-((3-(2-hydrazinyl-2-oxoethyl)phe-
nyl)diazenyl)benzyl)benzamide ((E)-1). A solution of 7 (0.032 g,
0.069 mmol) in anhydrous methanol (0.8 mL) was treated with
hydrazine monohydrate (33.5 µL, 0.69 mmol). The reaction was
left overnight under nitrogen before removal of the solvent under
vacuum to give an orange oil which was purified by flash
chromatography (MeOH with 10% NH4OH/CH2Cl2 gradient, 1:39
Photoresponsive libraries of this type are useful in that, despite
the nature of the guest, binding can be controlled by irradiation
at appropriate wavelengths. This shows the potential to, in one
system, both identify and synthesize novel host receptors and
take advantage of the optical properties of azobenzene to control
binding.
1
to 1:19 to 1:9) to afford the monomer 1 (65%). H NMR (CDCl3,
300 MHz): δ 7.88-7.77 (m, 6H), 7.58 (d, J ) 7.5 Hz, 1H),
7.50-7.36 (m, 5H), 5.39 (s, 1H), 4.74 (d, J ) 5.7 Hz, 2H), 3.63
(s, 2H), 3.30 (s, 6H). HRMS-ESI(+)+: m/z [M + Na] calcd for
C25H27N5NaO4 ) 484.1961, found ) 484.1958.
(Z)-3-(Dimethoxymethyl)-N-(3-((3-(2-hydrazinyl-2-oxoethyl)phe-
nyl)diazenyl)benzyl)benzamide ((Z)-1). (E)-1 was photolyzed at 365
nm. 1H NMR (CDCl3, 300 MHz): δ 7.81 (s, 1H), 7.73 (d, J ) 7.8
Hz, 1H), 7.53 (d, J ) 7.8 Hz, 1H), 7.38 (t, J ) 7.8 Hz, 1H),
7.14-7.01 (m, 3H), 6.75-6.58 (m, 5H), 5.71 (d, J ) 12 Hz, 2H,
NH2), 5.37 (s, 1H), 4.46 (d, J ) 5.7 Hz, 2H), 3.34 (s, 2H), 3.28 (s,
6H).
(26) (a) Bulos, F.; Roberts, S. L.; Furlan, R. L. E.; Sanders, J. K. M. Chem.
Commun. 2007, 3092–3093. (b) Ludlow, R. F.; Liu, J.; Li, H.; Roberts, S. L.;
Sanders, J. K. M.; Otto, S. Angew. Chem., Int. Ed. 2007, 46, 5762–5764.
(27) Gorostiza, P.; Volgraf, M.; Numano, R.; Szobota, S.; Trauner, D.; Isacoff,
E. Y. Proc. Nat. Acad. Sci. U.S.A. 2007, 104, 10865–10870.
(28) Boulegue, C.; Loweneck, M.; Renner, C.; Moroder, L. ChemBioChem
2007, 8, 591–594.
116 J. Org. Chem. Vol. 74, No. 1, 2009

Photoswitchable Dynamic Combinatorial Libraries
(S)-N-(3-Dimethoxymethylbenzoyl)prolinecarboxylic Acid Hy-
drazide (2). Monomer 2 was prepared according to literature
procedures,20 except the amide bond formation was accomplished
with HBTU/HOBt as the coupling reagent rather than EDC, as
described below. A solution of 0.059 g (0.302 mmol) of 3-car-
boxybenzaldehyde dimethoxy acetal, 0.041 g (0.302 mmol) of
HOBt, and 0.115 g (0.302 mmol) of HBTU in 1.5 mL of DMF
was added to 0.050 g (0.302 mmol) of L-proline methyl ester
hydrochloride. To the mixture was added 126 µL (0.906 mmol) of
triethylamine, and the solution was stirred for 1.5 h. After the
addition of brine, the product was extracted with EtOAc. The
organic phase was washed with sodium bicarbonate and dried over
MgSO4, and the solvent was evaporated to give 0.088 g (94%) of
a colorless oil. The hydrazinolysis reaction of the methyl ester was
carried out as previously reported to afford 2.
4 days. The reaction was monitored daily by LC-MS (3 µL
injections). Separations were performed using H2O-acetonitrile
gradients with 0.2% formic acid (t ) 0 min: 100% water, flow rate
1.0 mL/min; t ) 2 min: 75% water, flow rate 1.0 mL/min; t ) 3.5
min: 66% water, flow rate 1.0 mL/min; t ) 6.5 min: 57% water,
flow rate 1.5 mL/min; t ) 11 min: 30% water, flow rate 1.5 mL/
min), with the left column temperature set to 40 °C and the right
column temperature set to 50 °C to optimize separation. Multiple
wavelengths were used for analysis (220 nm, 254 nm, 280 nm,
and 320 nm). The cis and trans isomers were assigned based on
the strong absorbance of trans azobenzene at 320 nm, along with
the shift in the distribution of isomers upon isomerization.
(b) DCL with Monomers 1 and 2 or 1 and 3. The mixed library
was prepared by making a 1:1 solution of the building blocks (4
mM each) in CHCl3-DMSO (85:15 v/v) containing 100 mM of
TFA. The resulting solution was stirred at room temperature for at
least 4 days for equilibration. The reactions were monitored daily
by LC-MS as described above for the single component library.
In the case of templation studies, a second library was equilibrated
in the presence of 25 mM of a five-residue polyproline peptide.
Photoisomerization. Libraries were irradiated at 365 nm for 5
min in the dark with a Spectroline long wave UV pencil lamp (1,000
µW/cm2 of 365 nm radiation at 1 in.). Once irradiated, solutions
were continually stored in the dark at room temperature. LC-MS
analysis was continued on a daily basis, and light exposure was
minimized during analysis.
(S)-N-(4-Dimethoxymethylbenzoyl)prolinecarboxylic Acid Hy-
drazide (3). The synthesis of (S)-N-(4-dimethoxymethylbenzoyl)-
prolinecarboxylic acid hydrazide (3) was carried out using 4-car-
boxybenzaldehyde dimethoxy acetal and the same conditions as
those reported above for monomer 2.
Polyproline Peptide (Ac-Pro-Pro-Pro-Pro-Pro-NH2). The peptide
synthesis was performed on a Tetras Peptide Synthesizer using
Applied Biosystems PAL-PEG amide resin. The peptide was
synthesized on a 0.0215 mmol scale (50 mg resin). Coupling
reagents were HOBt/HBTU in DMF. The N-terminus was acylated
with a solution of 5% acetic anhydride and 6% 2,6-lutidine in DMF.
Cleavage was performed by hand with a cocktail of 95% TFA/
2.5% triisopropylsilane/2.5% H2O. The peptide was purified by
semipreparative reversed-phase HPLC on a C18 column at a flow
rate of 4 mL/min. The purification was achieved using a linear
gradient of A and B (A: 95% H2O/5% CH3CN with 0.1% TFA; B:
95% CH3CN/5% H2O with 0.1% TFA) and elution was monitored
at 214 nm. Once purified, the peptide was lyophilized and
characterized by ESI-MS. ESI-MS (+): m/z [M+H]+ calculated
for C27H40N6O6 ) 545.30, found ) 545.3.
Acknowledgment. We acknowledge the Army Research
Office and the Defense Threat Reduction Agency (DTRA) for
support (W911NF-06-1-0169 and HDTRA1-08-1-0045). We
also thank Professors Michael R. Gagne´ and James W. Jorgen-
son (UNC Chapel Hill) for helpful discussions.
Supporting Information Available: General experimental
details; 1H NMR spectra for (E)-1, (Z)-1, and compounds 2-7;
UV-vis spectra of (E)-1 and (Z)-1. This material is available
free of charge via the Internet at http://pubs.acs.org.
Dynamic Combinatorial Chemistry. (a) DCL with Monomer
1. The single-component library was prepared by making a 4 mM
building block solution of 1 in DMSO containing 100 mM of TFA.
The resulting solution equilibrated at room temperature for at least
JO801783W
J. Org. Chem. Vol. 74, No. 1, 2009 117